How Blossom is made

The short version

Yes, AI is involved, but only in one way, and it is probably not the one you are asking about. AI writes most of Blossom's code. Nothing inside the app sends what you write to an AI service. Blossom is ordinary code doing ordinary things.

That is explained properly below, along with what happened to the one feature that did use AI. If you would rather check than read, the code is public at github.com/elliegreydev/project-blossom.

AI writes most of the code

I use AI heavily to write Blossom. I am one person, I am not a developer by trade, and this app would not exist without it.

Being straight rather than letting you find out from the source: I do not read every line. What I do is decide what goes in and what stays out, and I have said no to plenty. The judgement about what Blossom should be, what it refuses to do, and where your data is allowed to go is mine.

That is easy to say, so here are three you can go and check. The waiting list has no progress bar, because a bar that fills up would be a lie about a queue that is not moving. Blossom keeps no waiting times of its own, only what you were told and when, because a number typed in during August is wrong by Christmas, and being wrong about a queue means somebody decides not to ring. Photo backup was designed in full and then not built, because doing it safely without a password meant inventing key management from scratch, and getting that wrong loses somebody their photos for good.

Each of those is written into the code with the reasoning beside it, so the decision cannot quietly get undone later. That is what the long comments are for.

What that means in practice is that you should not take my word for how it behaves, and you do not have to. The repository is public specifically so the privacy claims can be checked rather than believed. The parts that matter most have tests you can run yourself without a database or an account. Code written badly still cannot send somewhere it was never told to send.

If you read it and find something wrong, especially anything touching privacy, I would genuinely rather know. Open an issue on the repository, or email us.

Aurora is not AI

Aurora is the gentle nudge on your Home screen, and the name confuses people, so it is worth saying outright: it is not AI. Those suggestions are plain rules, worked out on your own device from what you have already told Blossom. Nothing you have written is sent anywhere to work one out, and they need no connection at all. If you turn sync on, the only thing that travels is a note of which nudge you have already seen, so the same one does not repeat on your other devices.

There used to be a second feature sharing that name, an optional chat that sent what you typed to Anthropic. It has been removed. It was the only part of Blossom that sent anything you wrote outside the UK, and that is no longer true of any part of the app.

What Blossom does with your data, now that this is the whole picture, is set out in the Privacy Policy, which is the complete version if this summary and that document ever disagree.

What Blossom is never allowed to do here

Blossom gives no dosing guidance, no information about where to buy anything, and never interprets a blood test result. Those are clinical, they are not ours to give, and no amount of asking changes it.

Blossom cannot diagnose, prescribe, monitor an emergency, or contact anyone on your behalf. Nothing in Blossom decides anything about you at all.

Why this page exists

Because people asked, and they were right to. An app that holds this kind of information about people who often have very good reasons to be careful should say plainly how it was built and what it does with what you write. A declaration you have to go digging for is not really a declaration.

Contact

Email support@projectblossom.net with anything about this, including if you think something here is wrong. Inside the app, Settings then Contact support reaches the same inbox.